Intelligent and Automatic Inspection, Reconstruction and Process Planning Methods for Remanufacturing and Repair

Abstract

Nowadays, it is critical to explore methods for dealing with worn and damaged components because of rising concerns over escalating emissions, resource depletion, and other environmental issues. Remanufacturing and repairing have been considered environmentally friendly manufacturing strategies to resist the increase in waste production. However, these processes are lacking automation support because compared to the manufacturing process, stochastic returns of used parts and their uncontrollable quality condition results in a high degree of uncertainty for the remanufacturing and repairing process. To fix these issues, the research proposed aims to automate and optimize the processes by improving the research areas in three perspectives: damage inspection (identification and localization), nominal volume reconstruction and process planning. The developed methodology features (a) an intelligent inspection to support automated classification and localization of damages from the end-of-life part; (b) an efficient geometric reconstruction of the damaged part to support the damaged volume extraction; (c) cost-driven and collision-free process planning that support hybrid remanufacturing processes. From the perspective of intelligent damage inspection, the RGB images and depth images are acquired by a depth camera. Then, the deep learning neural network segments and classify the damage. With that result, the spatial position of the damaged area is calculated by the integration of damage segments and the point clouds. In the damage volume extraction view, an efficient and precise nominal model reconstruction method is developed, which includes the processes of StepRANSAC surface fitting, SDM refitting, modelling, damage volume localization, and extraction. The result of reconstruction significantly facilitates the repairing process because it does not require prior information on the nominal CAD model. For hybrid remanufacturing, two hybrid manufacturing strategies are explored separately in developing the collision-free and cost-driven process planning. In both systems, the level of automation of hybrid processes is strongly increased. Finally, the conclusion, discussion and future work are followed. The proposed research can be extended to investigate damage quantification, freeform surface-based model reconstruction, process planning for complicated cases in future work.